Blog
What are we learning while building Inhold?
Inhold's blog has notes from building agent memory for hospital security review, written by the founder. Each post says what is built and what is not.
Inhold is being built. No customers yet. Repo goes public 9 Nov 2026.
What a hospital security reviewer asks about AI agent memory
The questions a hospital security reviewer asks about AI agent memory, and how Inhold answers each one today, with what is built and what is not.
31 Oct 2026
Why every audit row is chained to the one before it
Inhold hash-chains every audit row. Here is what the chain proves, what it cannot prove against a database admin, and how anchors close part of that gap.
31 Oct 2026
Agent memory can be PHI. Where should it live?
What an AI agent remembers about a patient can be PHI. Why Inhold keeps that memory in the hospital's own cloud, and what that does and does not settle.
31 Oct 2026
We run our own company's AI agents on Inhold
Inhold's own AI agents are Inhold principals with one purpose each. What the plugin enforces in code, what is switched on, and what is still a plan.
31 Oct 2026
What 30 red-team cases taught us
Inhold's 30 red-team cases try to cross purpose and identity boundaries in a seeded synthetic hospital. What each group checks and what the test asserts.
31 Oct 2026
RSS feed for the blog and the build log.